Work package number: WP3
Work package title: Long-term Continuous Geodetic Monitoring of Crustal Deformation
WP Leader: TUBITAK
Objectives
In this WP, long-term continuous monitoring of the crustal deformation will be investigated by exploiting the existing geodetic crustal deformation monitoring systems (Marmara Continuous GPS Network, with the complementary GPS surveys) (Task 1). Additionally, we propose to process SAR data, made available through the Supersites Initiatives archives, acquired by the old and new generation radar sensors, to compute the time series of the occurred and on-going surface displacements (Task 2). To this aim, two different advanced InSAR techniques, the SBAS and PSI ones, will be applied to C-, X- and L- band SAR data. Hence, the integration of the GPS, SBAS and PSI measurements (Task 3), with the contribution of seismological data, will allow us to map the dense spatial-temporal evolution of the present-day crustal deformation phenomena affecting the MARsite area. After the separation of the regional and local deformation processes, we will develop analytical and numerical modelling to define the seismic cycle and map the deformations on the secondary branches of the NAFZ (Task 4). While studying the ERS1/2 and ENVISAT radar data sets, we will update the algorithms and software tools for the future ESA GMES Sentinel-1 A and B satellites (Task 5) and we will be ready for the future. To increase the quality of advanced InSAR analysis, we will develop new approaches to reduce the atmospheric in-homogeneities at the time of acquisition of the different SAR images (Task 6). All efforts will be combined to better determine the 4D deformations in order to understand earthquake cycle processes, to develop probabilistic earthquake forecasting models and to constrain the seismic hazard models in the Marmara region.
Description of work
Task 1. Land-based continuous monitoring of crustal deformation
Interpretation of the data, from existing geodetic crustal deformation monitoring systems (Marmara Continuous GPS Network of TUBITAK-MAGNET, with the complementary TUBITAK GPS surveys) show that the Marmara region is subject to faulting, compaction induced subsidence, inflation and landslides, each of which process is posing a hazard to population and infrastructure. This is a crucial task to measure the tectonic strain accumulation across the Istanbul metropolitan area and western section of the 1999 Izmit rupture by combining the InSAR and GPS data. During the project, this task will supply the key geodetic ground control data to other task, based on the short- and long-term deformations in order to produce the hazard maps. MAGNET daily data flows to TUBITAK’s archive and merges with historical data, automatically. Using the daily updated archive, the GPS time series will be analysed to catch the short time deformation analysis, continuously. In addition, continuous time-series will be merged with survey data and the velocity maps will be obtained in semi-annual periods, in order to define long-term secular motions in detail.
Task 2. Exploitation of the SBAS and PSI algorithms for surface deformation analysis
2.1 IREA intends to apply the advanced version of the SBAS technique to X-band SAR data acquired by the new generation radar sensors, made available through the Supersites Initiatives archives. This will allow monitoring the temporal evolution of crustal deformation occurring in selected areas of the NAFZ via the generation of displacement velocity maps and deformation time-series.
2.2 BRGM proposes to process SAR data made available through the Supersites Initiatives archives, acquired by the archived C-band radar sensors to retrieve time-series of surface displacement on selected areas of the NAFZ. This will allow us to map the spatial-temporal evolution of the present-day crustal deformation phenomena affecting the MARsite Area with high level of temporal/spatial details. The goal is to highlight the long-term behaviour of active faults and eventual interactions between structures. Complementarily, where possible on selected areas -nominally on secondary branches of the NAFZ- we also propose to use the archived example L-Band data to demonstrate the advantages of L-band.
2.3 INGV will define in agreement with the other teams, selected areas over which start a detailed monitoring using the X-band COSMO-SkyMed constellation, with a revisit time of 4 or 8 days in ascending and descending geometries. INGV will also process the COSMO data using the SBAS or PSI techniques, depending on the area. . The high frequency of InSAR monitoring is expected to provide new information on possible deformation transients in the pre-seismic phase, while in case of seismic event the 4D deformation maps will monitor the evolution of the post-seismic strain diffusion. Moreover, the high-resolution deformation maps provided by COSMO are needed in Task 4 to separate the regional and local deformation processes.
Task 3. Integration and harmonization of InSAR, GPS and seismic data
Task 1 of WP2, Tasks 1 and 2 of WP3 and Tasks 1 and 2 of WP5 will be the main data sources for this task. In addition, the PSI products of TERRAFIRMA project will be used. In the framework of TERRAFIRMA the European Space agency (ESA) made available the whole SAR database (ERS-1-2 and Envisat) to be used applying PSI in order to obtain surface velocity maps and time series all over the Marmara Region area. In particular, the available PSI products of TERRAFIRMA cover the time interval 1992-2009. This time interval will be extended with new TerraSAR-X and COSMO-SkyMed data sets. The results will be validated with measurements in Task 1 of WP3 and other in situ data made available in other WPs.
The short-time revisit capability of COSMO-SkyMed data is extremely important when studying the theoretically predicted precursory phenomena to earthquake preparation (dilatancy), and the various processes occurring in the post-seismic phase: dilatancy recovery, pore pressure readjustments, afterslip and visco-elastic relaxation. The integration of high resolution InSAR deformation maps with the precise CGPS measurements is the only possible way to fully appreciate the patterns of these elusive signals, whose understanding is crucial to verify (or develop) the theories describing the seismic cycle.
In conclusion, under the contribution of seismological data sets, we will focus to integration and correlation of different data sources, for different earthquake data sets in the past and future.
INGV In the framework of the European project Terrafirma the European Space agency (ESA) made available the whole SAR database (ERS-1-2 and Envisat) to be used applying PSI in order to obtain surface velocity maps and time series all over the Marmara Region area. In particular, the available PSI products of TERRAFIRMA cover the time interval 1992-2009. This time interval will be extended with new TerraSAR-X and COSMO-SkyMed data sets. The results will be validated with measurements in Task1 of WP3 and other in situ data made available in WPs, and will be used for the modelling activities and CFF estimates in Task 4.
We will carry out the joint analysis of CGPS data and DInSAR time-series, to provide more accurate and cross-validated ground velocity maps. We will use the CGPS data to constrain the deformation components at long spatial wavelengths. Using the different information content of CGPS and DInSAR data we will model the effects of possible error sources due to atmosphere, topography, orbital biases.
Task 4. Separation of the regional and local deformation processes and modelling
GPS and InSAR data commonly show various interfingered deformation processes. Separation of the regional and local deformation processes is required to further utilize data for kinematic and physical models. Using decomposition approaches such as those based on singular values, we propose to identify and separate the overlapping deformation signals. The goal is to identify dominant signatures in the data, which might be visually hidden due to their temporal and spatial scale. Furthermore, the task is to use these signals separately for quantitative analysis. This will be done by inversion methods that will be further developed to model both the original data and the decomposed signatures. The aim is to improve understanding of both local processes and regional scale processes. Local processes might be land compaction or landslides. Regional scale deformation processes might be the inter-seismic steady-state plate motion combined with co-seismic and transient deformation processes that have happened in the past, taking the full deformation time series and herewith time dependent rheological complexities into account.
In particular, modelling of the deformation processes is foreseen in the following ways; (a) elastic dislocation block-models with the aim to study microplate kinematics in the framework of major plate convergence, and active strain build-up at block-bounding faults. Kinematically consistent elastic block-model will be used to infer the pattern of fault-coupling on the plate-boundary faults, by a constrained inversion of GPS and InSAR velocity maps. (b) Time-series deformation data (from Multitemporal InSAR and CGPS) will be modelled using both analytical and numerical modelling techniques. Based on principal component analysis the space-time evolution of slip on fault planes is to be investigated during both the interseismic and post-seismic phases (for both archived and new incoming data streams). (c) Modelling of time-series data shall allow investigation of different rheological behaviours in the body and the fault zone, such as those associated with creep, visco- and poro-elasticity and plastic deformations. (d) Modelling deformation data together with double integration of accelerometer data (see WP5), and finally (e) models shall allow to analyse the fault interaction with the Coulomb Failure Function (CFF), i.e. co-seismic and interseismic perturbations to the regional stress field.
As a final goal, the developed models shall be investigated with respect to microseismological data (see WP2 and WP4) to detect branches of the NAFZ, and to evaluate the power of such improved data handling for probabilistic earthquake forecasting.
Task 5. Extension and the transition into the new (GMES) satellite constellation and data for advanced InSAR analysis
The future ESA GMES Sentinel-1 A and B satellites will represent an unprecedented source of regular, consistent and frequent SAR data, of high interest for any application that calls for continuous monitoring of small terrain displacements. As soon as the constellation will be operational, a continuous coverage will be guaranteed, with one acquisition every 6 days with characteristics suitable for interferometric combination. This feature has been obtained by exploiting a new acquisition modality (TOPS) that, implementing a burst-mode and scanning geometry, allows covering very large areas while worsening some geometric resolution in one (azimuth) of the two directions.
The availability of Sentinel-1 data will also cover the gap after the change of orbit of the ENVISAT satellite that hindered the possibility to continue the formation of displacement time series over long-time intervals. The new Sentinel-1 acquisition modality, while very interesting, calls for a significant update of the algorithms and software tools that are exploited during advanced (PS + SBAS) InSAR analysis of long SAR data time series. The aim of this Task is first to update an existing operational processing chain for advanced InSAR analysis (the SARscape®Interferometric Stacking module), currently based on stripmap, spotlight and/or ScanSAR acquisitions, to also support the Sentinel-1 TOPS (Interferometric Wide Swath) acquisition mode. The launch of the first Sentinel-1 platform (Sentinel-1A) is currently scheduled for May 2013; it is hence foreseen that some of the developments will be performed with simulated data, and then with not fully-calibrated data originating from the mission’s CAL/VAL campaign, when available. The updated processing chain will be then validated and exploited for processing new data obtained from the Sentinel-1 operational phase, to start building new displacement time series.
Task 6. Integrating a few independent sources for atmospheric artefacts reduction (MERIS, MODIS, OSCAR from JPL, GPS) into PSINSAR and SBAS analysis
Atmospheric in-homogeneities at the time of acquisition of the different SAR images that are combined together to perform advanced InSAR analysis are a significant source of artefacts (the so-called Atmospheric Phase Screen) and distortions, that ad-hoc filtering aims to minimize to increase the final accuracy of the displacement measurements.
The typical approach of these filtering algorithms relies on different expected temporal and spatial distribution of the APS respect to the displacement signal to be measured. In particular, it is expected that the APS is (for satellite acquisitions separated of one or more days) temporally uncorrelated (high-pass signal), while the deformation signal has a significant temporal correlation (low-pass signal). This assumption is, of course, risky, in particular in case of abrupt events like earthquakes, where discontinuities of the measured displacement are often directly to be related to real deformations and not to artefacts.
The availability of external, independent sources to characterize, estimate and as far as possible subtract APS components is hence a key issue to minimize their impact on the final measurements accuracy and not to risk to mix the atmospheric and displacement signals in what is subtracted from the original data. In this case three main types of sources can be considered: multi-spectral imagery (e.g. MERIS and MODIS sensors), weather forecast (e.g. WRF, ECMWF) models and GNSS (e.g. GPS) measurements; these systems can be exploited to estimate Zenith Path Delay layers at the time of acquisition of the SAR images and compensate for their temporal difference.
One goal of this work-package is to build a set of software bridges between the existing PS and SBAS SARscape® processing chain and ZPD layers obtained from the different cited sources. In particular, one example of such a data source will be the OSCAR system of JPL that allows dedicated software clients to request and obtain ZPD information that is derived from combined MODIS and ECMWF data. Another source of particular interest for ENVISAT ASAR data is the MERIS instrument. This sensor can acquire multi-spectral data at the same time and on the same area of ASAR, providing as standard product a water vapour layer that can be simply converted into ZPD estimation.
The ZPD data will be integrated within the SARscape advanced InSAR processing chain to optimize the APS filtering stages, minimizing artefacts due to possibly wrong assumptions.
The extended processing chain will then be exploited to generate new versions of displacement time series over the test site area, allowing to analyse and to quantify the improvements that can be obtained with this combined approach.